Intelligent Systems for Medical Diagnostics with the Detection of Diabetic Retinopathy at Reduced Entropy

Diabetes Mellitus (DM) is primarily defined by hyperglycemia, polyuria, and polyphagia and as a result of a complex interaction of hereditary and environmental variables, it has developed into a severechronic metabolic condition. Numerous diabetes problems might arise from uncontrolled high blood sugar. Long-term diabetes causes severe complications, some of which are fatal. Everywhere in the world, the prevalence of diabetes in patients is increasing at epidemic rates. Every year, diabetes and related disorders consume a sizable percentage of the national health budget. Several risk factors influence the etiopathogenesis of the disease and the emergence of the epidemic. The untreatable condition of diabetes may be managed by maintaining self-care in daily life, providing appropriate education about diabetes, and making significant advancements in knowledge, attitudes, skills, and management. Diabetes should be diagnosed as soon as possible because it can lead to several illnesses, including kidney failure, stroke, blindness, heart attacks, and lower limb amputation. This study aims to use relevant variables, create a prediction algorithm using machine learning, and choose the best classifier to produce results that are as close to clinical outcomes as possible with reduced entropy. The proposed approach is on selecting the characteristics that ail in the early detection of Diabetes Miletus utilizing Predictive analysis. The computational techniques K-Nearest Neighbour (KNN) and Decision Tree (DT) have been employed to detect DM at an early stage. The KNN outperforms with the best performance when evaluated against different performance metrics.